Naval Command,Control and Ocean San Diego. CASurveillance Center RDT&E Division 92152-5001

ADaA28 2 369

Technical Report 1646

May 1994

UltrawidebandShipboard ElectroopticElectromagneticEnvironment Monitoring

S. A. Pappert DTICM. H. Berry A ELECTES. M. Hart JUL25199 4

R. J. Orazi m G DL. B. Koyama NJS.T.Li

94-22723

A ii M l Aproved for public rele ase; distribuion is un blled

Ivord

Technical Report 1646May 1994

Ultrawideband Shipboard ElectroopticElectromagnetic Environment Monitoring

Accesion For

S. A. Pappert NTIS CRA&MDTIC TAB

M. H. Berry Unannounced 0S. M . Hart Justification .................... .

R. J. OraziL.B .K ya aBy ......................... .L. B. Koyama Dist. ibution I

Availability Codes

Avail and orDist Special

NAVAL COMMAND, CONTROL ANDOCEAN SURVEILLANCE CENTER

RDT&E DIVISIONSan Diego, California 92152-5001

K. E. EVANS, CAFiT USN R. T. SHEARERCommanding Officer Executive Director

ADMINISTRATIVE INFORMATION

This report is submitted in partial fulfillment of Milestones 3, 4, 5, and 6, Task Num-ber 3 (Electrooptic Electromagnetic Monitoring) of the Electromagnetic CompatibilityProject (RH21C13) of the Surface Ship Technology Area Program (SC1A/PE0602121N).The work described herein was sponsored by the Office of Naval Research (ONR 334)under contract N0016791WX10040. The work was performed by the Naval Command,Control and Ocean Surveillance Center (NRaD Codes 555, 822, 843, and 753).

Released by Under authority ofMatthew N. McLandrich, Head Dr. Howard E. Rast, HeadOptical Electronics Branch Solid State Electronics

Division

SM

EXECUTIVE SUMMARY

OBJECTIVE

1. Design, construct, package, test, and evaluate the best available wideband photonicelectromagnetic field probe for shipboard electromagnetic environment (EME) moni-toring.

2. Develop engineering solutions to problems encountered in Phase I testing of a first-generation breadboarded sensor. Investigate and develop alternative sensor configura-tions that can lead to improved system performance.

3. Develop requirements and a test plan for the shipboard testing of a 2-MHz to 18-GHzprototype system.

RESULTS

1. Broadband electromagnetic (EM) field detection using antenna-coupled fiber opticlinks has been successfully demonstrated in the 2-MHz to 18-GHz frequency range.This remote sensing system is potentially operable from 2 MHz to 50 GHz and can bepackaged into small, lightweight units. Both amplitude and frequency information ofmultiple radio frequency and microwave signals have been simultaneously monitoredwith this wideband optical system. A root mean square electric field sensitivity ofapproximately 1 gV/m and a spurious free dynamic range of 109 dB/Hz213 have beendemonstrated with the 18-GHz externally modulated system.

2. Semiconductor optical waveguide modulators have been developed as an alternativeto lithium niobate Mach-Zehnder modulators for this application. Systems employingthe less mature semiconductor optical modulators have shown greater modulationefficiency for a given bandwidth at the expense of added optical insertion loss thanlithium niobate-based systems.

3. Remote modulator bias control techniques have been investigated and a system basedon the active control of the modulator direct current bias to compensate for bias posi-tion drift has been developed. This system utilizes a remote power-by-light approachin which a low-frequency optical control signal is sent to the modulator via an opticalfiber and is remotely monitored to properly bias the optical modulator. This systemhas been successfully demonstrated in the 10'C to 90°C temperature range.

4. A remote optical polarization controller has been successfully demonstrated, whicheliminated the need for expensive polarization-maintaining fiber. The controller hasan optical insertion loss of 4.5-dB, resulting in a 9-dB penalty in detection sensitivity.Recent improvements in liquid crystal technology have resulted in polarization con-troller units with

high-dielectric-constant substrate. A third approach has been to use a sectoralizedloaded monopole antenna. A fourth approach is to suffer the gain penalty and use anelectrically short dipole antenna. Each technique has its advantages and disadvan-tages, although the use of an electrically short dipole antenna is the simplestapproach, and results indicate that acceptable 2-MHz to 500-MHz electric field sensi-tivities can be obtained using this approach.

6. A preliminary test plan for the FY 94 shipboard demonstration aboard an LSD-41class surface ship is outlined. Preliminary results indicate that as few as three topsidemonitoring locations are required to obtain the electromagnetic signature of the ship.

RECOMMENDATIONS

I. Prepare the 2-MHz to 18-GHz prototype electrooptic EME system for a shipboarddemonstration test to occur in FY 94. Continue developing the semiconductor opticalwaveguide modulators to eventually replace the presently used lithium niobate modu-lators. Extend fiber optic link operation to 50 GHz using both semiconductor andlithium niobate modulators and compare their performance.

2. Develop and optimize radio frequency and microwave probe designs for continuous2-MHz to 50-GHz frequency coverage with compact, lightweight antennas.

3. Demonstrate EM field detection from 2 MHz to 50 GHz with a single fiber optic linkand as few antennas as possible.

iv

CONTENTS

EXECUTIVE SUM MARY ................................................... iii

ABBREVIATIONS ......................................................... ix

1.0 INTRODUCTION ..................................................... 1

1.1 TECHNICAL OBJECTIVE AND APPROACH ........................ 1

1.2 BACKGROUND ................................................. 2

2.0 SHIPBOARD EO ENE MONITORING .................................... 4

2.1 DESCRIPTION OF EXTERNALLY MODULATED EO EMEMONITORING SYSTEM .......................................... 4

2.1.1 Lithium Niobate Optical Waveguide Modulators ................. 6

2.1.2 1H-V Semiconductor Optical Waveguide Modulators .............. 7

2.2 ANALYSIS OF EO EME MONITORING SYSTEM .................... 10

2.3 PERFORMANCE OF FIBER OPTIC LINK ........................... 17

2.4 ELECTROMAGNETIC FIELD MEASUREMENTS IN THE2- to 18-GHz FREQUENCY RANGE USING AN EO EMEMONITORING SYSTEM .......................................... 20

2.4.1 System Configuration ....................................... 21

2.4.2 Results .................................................. 22

3.0 IMPROVEMENTS TO EO EME MONITORING SYSTEM .................... 28

3.1 IMPROVEMENTS TO FIBER OPTIC LINK .......................... 28

3.1.1 Ultrawideband Optical Modulators ............................ 28

3.1.2 Optically Powered and Controlled Remote Fiber Optic Links ....... 30

3.1.3 Remote Optical Polarization Control of EM Field Sensor ........... 37

3.2 IMPROVEMENTS TO WJDBAND RF PROBE ........................ 40

3.2.1 Electrically Short Dipole or Monopole Antenna .................. 41

3.2.2 Enlarged Spiral Antenna .................................... 42

3.2.3 Sectorialized Monopole Antenna .............................. 42

3.2.4 High-Dielectric Spiral Antenna ............................... 42

v

4.0 SHIPBOARD DEMONSTRATION TEST PLAN ............................ 47

4.1 SHIP CLASS SELECTION ........................................ 47

4.2 SHIPBOARD EME DETERMINATION .............................. 52

4.3 SHIPBOARD TEST AND EVALUATION PROCEDURES ............... 55

5.0 CONCLUSIONS ...................................................... 56

6.0 RECOMMENDATIONS ................................................ 58

7.0 REFERENCES ........................................................ 59

FIGURES

2-1. Schematic diagram of an antenna-coupled externally modulated EO EMEm onitoring system ..................................................... 5

2-2. Optical powering scheme using GaAs photovoltaic cells for the EO EMEm onitoring system ..................................................... 5

2-3. Triplexing scheme to combine RF information from 2 MHz through 50 GHz ....... 6

2-4. Schematic of lithium niobate MZ interferometric optical modulator .............. 7

2-5. Structure of LPE-grown 1.32-pm InGaAsP FKE optical modulator ............... 9

2-6. Structure of MOCVD-grown 1.52-1&m InGaAs/InP multiple quantum wellQCSE optical modulator ................................................ 9

2-7. Relative transmission versus applied bias curves for the 1.32-pam FKE waveguidemodulator, the 1.52-pm QCSE waveguide modulator, and the 1.32-tpm lithiumniobate MZ waveguide modulator ......................................... 11

2-8. Experimental values for Aa versus applied bias for the 1.32-pm FKE and1.52-pm QCSE waveguide modulators ..................................... 12

2-9. Optical receiver output power versus modulation index, including harmonicsand intermodulation products for the 1.32 -pm MZ modulator ................... 13

2-10. Optical receiver output power versus modulation index, including harmonicsand intermodulation products for the 1.32-pm FKE modulator ................... 13

2-11. Optical receiver output power versus modulation index, including harmonicsand intermodulation products for the 1.32-prm QCSE modulator ................. 14

2-12. Frequency response curve for 1.32-pm LPE-grown FKE modulator .............. 16

2-13. Plot of m2/(S/N), where m is the modulation index, in a 1-Hz bandwidthversus optical detector photocurrent for different laser RIN values ............... 17

2-14. Schematic diagram of best available DC to 18-GHz externally modulated fiberoptic link ............................................................. 18

vi

2-15. Modulation frequency response of wideband lithium niobate optical modulator ..... 18

2-16. Externally modulated fiber optic link RF insertion loss versus frequencyin the 50-MHz to 18-GHz range .......................................... 19

2-17. Diagram of the externally modulated fiber optic link loss budget ................. 19

2-18. Total noise power (solid line) and experimental data (circles) versus opticaldetector photocurrent for the externally modulated fiber optic link ............... 20

2-19. RF insertion loss versus frequency for a DC to 500-MHz fiber optic link usinga Va = 4 V optical modulator ............................................. 21

2-20. Schematic diagram of the antenna-coupled externally modulated fiber opticlink for EM field detection ............................................... 22

2-21. Modulation frequency response from 2 to 18 GHz of the externally modulatedfiber optic link and spiral antenna ......................................... 23

2-22. Plot of m2/(S/N), where m is the modulation index, due to total link noise(solid line), along with experimental data points (circles) ....................... 24

2-23. Plot of experimental (circles) and theoretical detected power for fundamental(solid line) and third-order intermodulation products (broken line) versuspower applied to the IOM ............................................... 24

2-24. EM field detection system output versus incident RMS electric field level ......... 26

2-25. Measured RMS electric field levels versus calculated RMS electric field levels ..... 27

3-1. Fiber optic link transfer function .......................................... 29

3-2. Schematic of optical powering system and modulator bias control circuitry ........ 31

3-3. Photocell output electrical voltage as a function of series resistance for 50-mWinput optical power ..................................................... 31

3-4. Photocell output voltage versus input optical power using a 2.7-kQ seriesresistance ............................................................ 32

3-5. Fundamental and spurious signal levels as a function of phase bias error for theminimum detectable modulation index ..................................... 33

3-6. Effect of 1-degree modulator phase bias error on the 30-500-MHz fiberoptic link ............................................................. 34

3-7. Modulator phase bias tolerance as a function of receiver bandwidth for the500-M Hz fiber optic link ................................................ 35

3-8. Modulator phase bias drift as a function of temperature for proton-exchangedand titanium-indiffused lithium niobate optical modulators ..................... 35

3-9. Computer-controlled modulator phase bias stabilization results as a function oftemperature for both low- and high-frequency fiber optic links .................. 36

vii

3-10. Remote polarization control setup using liquid-crystal phase retarders ............ 38

3-11. Fiber optic link RF insertion loss with optimum polarization alignment ........... 39

3-12. Fiber optic link RF insertion loss with worst-case polarization alignment .......... 39

3-13. Fiber optic link RF insertion loss with optimum liquid-crystal phase-retarderalignm ent ............................................................ 40

3-14. Gain versus frequency for the Watkins-Johnson two-arm spiral antenna ........... 41

3-15. Antenna output power into 50-9 load versus frequency for a bare andresistively loaded 15-cm dipole with 0.7-V/m RMS electric field excitation ........ 43

3-16. Minimum detectable RMS electric field level versus frequency for a baredipole, assuming a 40-dB system noise figure and a 30-kHz receier bandwidth ...... 43

3-17. Return loss (S11) versus frequency for 3-inch, 1.5-turn spiral antennasfabricated on relative dielectric constant 2.2 (squares) and 80 (crosses)substrate m aterial ...................................................... 45

3-18. Return loss (S11) versus frequency for 6-inch, 1.5-tumm spiral antennasfabricated on relative dielectric constant 2.2 (squares) and 80 (crosses)substrate m aterial ...................................................... 45

3-19. Electrical return loss (S11) for 6-inch spiral antennas on 2.2 dielectricconstant substrates with 1 (crosses), 1.5 (squares), and 2 (diamonds) turns ......... 46

3-20. Electrical return loss (S11) versus frequency for a 12-inch, 6-turn spiralantenna on 2.2 dielectric constant substrate .................................. 46

4-1. Bow of USS Germantown (LSD-42) ....................................... 48

4-2. Stern of USS Germantown (LSD-42) ...................................... 48

4-3. Forty-five degrees from bow, port view of USS Germantown (LSD-42) ........... 49

4-4. Forty-five degrees for stem, port view of USS Germantown (LSD-42) ............ 49

4-5. Broadside, port view of USS Germantown (LSD-42) .......................... 50

4-6. Broadside, starboard view of USS Germantown (LSD-42) ...................... 50

4-7. Candidate topside measurement sites aboard LSD 41 class ship .................. 54

TABLES

2-1. Externally modulated EO EME monitoring system electric field level detectionranges at 2.25, 9.52, and 16.0 GHz ........................................ 27

4-1. LSD-41 class antenna characteristics ....................................... 51

viii

ABBREVIATIONS

BW bandwidthEA electroabsorptionEHF extremely high frequencyEM electromagneticEMC electromagnetic compatibilityEMCON electromagnetic controlENE electromagnetic environmentEMI electromagnetic interferenceEO electroopticFKE Franz-Keldysh effectHF high frequencyIOM integrated optical modulatorLNA low-noise amplifierLPE liquid phase epitaxyMF multimode fiberMOCVD metalorganic chemical vapor depositionMZ Mach-ZehnderNF noise figurePBL power-by-lightPMF polarization-maintaining fiberQCSE quantum-confined Stark effectQMS quality monitoring systemQW quantum wellRF radio frequencyRIN relative intensity noiseRMS root mean squareSATCOM satellite communicationSFDR spurious free dynamic rangeSL superlatticeSMF single-mode fiberS/N signal-to-noise ratioUHF ultrahigh frequencyVHF very high frequencyVSWR voltage standing wave ratio

ix

1.0 INTRODUCTION

The shipboard electromagnetic environment (EME) routinely consists of high-level on-shipemissions extending from the high-frequency (HF) band into the extremely high frequency(EHF) band. Rapid advances in the areas of radar, electronic warfare (EW), and communica-tions technologies are making this EME more complex and at the same time more difficult tomanage. The Navy has an increasing need to monitor these emissions and verify electromag-netic control (EMCON) status as well as alert personnel of hazardous emission levels. Toaddress this need, the Navy is seeking affordable, broadband shipboard EME monitoring probesand systems that are as nonperturbing as possible.

In response to this shipboard EME monitoring requirement, photonic techniques are beingdeveloped in association with wideband radio frequency (RF) probes for use in EME monitoringsystems. The emerging photonic technologies will enable the Navy to collect and transmitbroadband shipboard EME information for processing at a remote site while minimizing theoverall system intrusiveness. Recent advances of high-speed fiber optic and electrooptic (EO)components has made this EME monitoring system approach viable and will provide shipoperators with a valuable new command, control, communication, computer, and intelligence(C41) capability. This status report reviews past EO EME monitoring system work and presentsresults from current efforts geared toward making this technology feasible. Requirements aswell as test and evaluation plans for the shipboard demonstration to be conducted under thisprogram in FY 94 are also presented.

1.1 TECHNICAL OBJECTIVE AND APPROACH

The principal objective of the Office of Naval Research-sponsored Electromagnetic Compat-ibility (EMC) Project is to minimize exterior electromagnetic interference (EMI) problemsduring the entire life cycle of Navy ships. An additional objective is the reduction or control ofthe RF emission signature of Navy surface combatants. This involves the task of controllingshipboard RF emissions by remote monitoring of the entire shipboard RF spectrum to determine,verify, and enforce the EMCON status.

Task Number 3 of the EMC Project develops a broadband, large-dynamic-range EM fieldprobe that can be packaged into a small, lightweight system. The Task couples EO devices towideband RF probes to provide a low-cost system that can monitor EM emission from the ship.This system is potentially useful from 2 MHz to 50 GHz, where most shipboard emission occurs.In addition to EMCON monitoring, the potential benefits or usages of this system will includethe detection of RF hazards levels at key locations on the ship weather decks, the qualitymonitoring of communication emitters (QMS), and real-time automated frequency management.

The technical approach of this task is to build upon current photonic technology and bridgethe gap where technology deficiencies exist to demonstrate a 2-MHz to 50-GHz EO EM fieldsensor. Photonic and wideband antenna component performances are being extended wherenecessary to meet the shipboard application requirements. The task will culminate in a ship-board demonstration of a prototype EO field sensor at which time the effort will be transitionedfor further engineering development.

This four-year R&D task (FY 91 to FY 94) is being performed in three phases. Phase I wasstarted and completed in FY 91; Phase II was started in FY 92 and will be completed in FY 94;

1

and Phase M11 is to begin in FY 93 and conclude in FY 94. Phase I (Milestones 1 and 2) con-sisted of assessing the present state-of-the-art of EO field sensors and determining their suitabil-ity for shipboard use. Phase II (Milestones 3, 4, 5, and 6) of this effort deals with solvingproblems determined from Phase I breadboard testing of the candidate EO EME monitoringsystem, developing alternative sensor schemes to improve system performance and extendfrequency coverage, optimizing the RF probe design for efficient wideband coverage, anddetermining design guidelines for a concept demonstration of the EO sensor. Phase HI of thiswork will consist of designing, constructing, and packaging the best available EO field sensorinto a prototype demonstration system. System performance will be measured and the environ-mental ruggedness of the prototype system will be assessed. Shipboard testing and a subsequentfinal report will finish the advanced development work assosciated with this effort.

This report describes the work performed in Phase HI (Milestones 3, 4, 5, and 6) of this Task.

1.2 BACKGROUND

Electromagnetic field sensing using EO and fiber optic techniques has been of interest to theNavy for some time. Previous work in this area has considered antenna-coupled lithium niobatebulk crystal and waveguide modulators, where high sensitivities have been attained for systembandwidths below 1 GHz. 1-3] This class of field sensor, which uses an optical intensitymodulator for the RF electrical-to-optical conversion, is referred to as an externally modulatedfield monitoring system. An alternative EO field sensing approacl has focused on the directcurrent modulation of an antenna-coupled high-speed injection laser diode.[451 This is referredto as a directly modulated field monitoring system. Both approaches use fiber optic links totransmit the EME information to a remote processing site. Directly modulated short-haul fiberoptic systems possess a simpler design and are easier to implement, whereas externally modu-lated systems have been shown to be more sensitive and possess larger 3-dB bandwidths. [6.71

Phase I work of this project demonstrated that the externally modulated EO EME monitoringsystem will ultimately outperform the directly modulated system and better meet shipboardEMCON monitoring system requirements. This conclusion was based on the superior noisecharacteristics of high-power solid-state lasers- (which can be used for externally modulatedsystems) compared to those of injection laser diodes and on the superior high-frequencymodulation characteristics of external waveguide modulators.

As stated above, Phase I[ of this program began in FY 92 and involved determining solutionsto problems encountered in Phase I and extending the performance capabilities of the selectedEM field monitoring system. Anechoic chamber testing of an externally modulated EO EMEsystem has been performed in the 2- to 18-GHz frequency range. A root mean square (RMS)electric field sensitivity of 15 RiV/m and a spurious free dynamic range (SFDR) of 102 dB in a1-Hz resolution bandwidth have been measured with this 2- to 18-GHz field detection system.These chamber results indicate that the externally modulated EO EME system can be useful forthe proposed shipboard applications. Greater sensitivity is expected by increasing the opticalpower of the laser and by improving the optical modulator's RF efficiency. Both these modifica-tions have been implemented. Laboratory testing of these refined EO sensor systems has beenperformed, and an increased sensitivity and dynamic range have been attained over the prior

*Available from Amoco Laser Co.

2

anechoic chamber results. A broadband fiber optic link noise figure of < 30 dB has beenmeasured, which translates into a 1-Hz RMS electric field senstivity of approximately 1 jiV/m.

A number of other system modifications have been investigated in an attempt to work outall the problem areas associated with this sensor. These Phase II modifications include thedevelopment of modified optical fiber cable assemblies to minimize polarization drift, theincorporation of polarization-insensitive semiconductor modulator designs and polarization-control devices to eliminate the need for polarization-maintaining fiber, and the operation oftemperature-insensitive fiber optic links.

Developing more efficient optical modulators and extending their frequency range to 50 GHzis an ongoing thrust of the Phase II work, with both in-house and industrial efforts. Theconventional design for externally modulated EO EME monitoring systems includes lithiumniobate optical waveguide modulators. Although lithium niobate waveguide modulators arerelatively mature and available with bandwidths to 18 GHz,* these broadband devices havecomplicated traveling-wave electrode designs and possess half-wave voltages in the 10-V range.An alternative to lithium niobate waveguide modulators for the externally moduated EO EMEsystem are III-V semiconductor waveguide modulators. Both interferometric and absorptionmodulators are possible,18,91 a'hough emphasis in this report has been on semiconductorelectroabsorption (EA) waveguide modulators. Both quantum-confined Stark effect[1°0 (QCSE)and Franz-Keldysh effect['1I (FKE) EA devices operating at 1.52 Itm and 1.32 gtm, respectively,are included. Both of these devices rely for operation on material absorption coefficient changeswith an applied electric field. Fiber optic link sensitivity. spurious free dynamic range (SFDR),and bandwidth predictions using these EA devices based on theoretical and experimental resultsare presented and compared to those of lithium niobate-based fiber optic links.

A final photonic probe design for the FY 94 shipboard demonstration has been completed.Fiber optic link implementations include a power-by-light (PBL) system to optically power themodulator as well as a modulator bias-control circuit to automatically adjust for environmentalchanges. These two system insertions improved the usefulness of the EO field sensor, and theresults of testing these system modifications will be presented.

Phase II antenna development efforts include the development of a compact HF/very highfrequency (VHF) antenna as well as a wideband spiral antenna useable to 50 GHz. Advancedantenna developmental efforts include the development of compact HF/VHF antennas. Oneapproach being pursued is the development of a small-aperture, high-dielectric spiral antenna,which should provide useable gain down into the HF band. Other low-frequency antennacandidates include an enlarged spiral antenna, a sectoralized monopole antenna, and an electri-cally short dipole antenna. Results using these HF/VHF/ultrahigh frequency (UHF) antennastructures are presented. A miniature spiral antenna design with gain to 50 GHz is also beingdeveloped and will be discussed. These antenna R&D efforts will continue into FY 94.

An initial test plan for the shipboard demonstration of the photonic EM field probe has beendeveloped. The prototype sensor will be placed at selected topside sites of an LSD-41 classsurface ship and controlled emissions will be recorded. Detailed testing procedures will bedescribed.

*Available from United Technologies Photonics.

3

2.0 SHIPBOARD EO EME MONITORING

In this section, an externally modulated EO EME monitoring system is described andlaboratory results on a number of breadboarded systems are presented and compared withsimulations. The photonic link as well as the wideband RF probe will also be discussed.

2.1 DESCRIPTION OF EXTERNALLY MODULATED EO EME MONITORINGSYSTEM

A schematic diagram of an antenna-coupled externally modulated EO EME monitoringsystem is shown in figure 2-1. Its primary components include a high-power, low-noise,polarized optical source, polarization-maintaining single-mode optical fiber (PMF) for theuplink, an antenna-coupled optical waveguide modulator, standard single-mode fiber (SMF) forthe downlink, a high-speed p-i-n photodiode, and a signal processor. The signal processor willnormally consist of a wideband spectrum analyzer and a central processing unit. Low-noisepreamplifiers are not included in the EO EME monitoring system configuration and are notconsidered necessary for this relatively high signal level application. The antenna and opticalmodulator are positioned at a selected point about the ship for EM field detection, whileamplitude and frequency information is remotely received and processed. The intent of thisgeometry or configuration is to minimize the perturbation of the EM field by eliminating anyelectrical transmission lines between the probe and receiving station. If passive modulatorbiasing is not feasible, electrical powering or biasing of the optical modulator can be achievedoptically via a high-power laser diode, a multimode optical fiber, and photovoltaic cells foroptical-to-electrical power conversion. This optical powering scheme is generically shown infigure 2-2 and its implementation is discussed in Section 3.1.

Initial laboratory prototypes of the EO EME monitoring system operate from 2 to 18 GHzand use a wideband spiral antenna which possesses a 50 Q output impedance and a minimumgain of 0 dBi across this band. A system is presently being constructed for shipboard testing,which is operable from 2 MHz to 18 GHz. This prototype system has been designed to operatewith a number of different optical modulator types that are being developed as part of this R&Deffort and that are described below. A four-fiber, all dielectric optical cable has been designedand fabricated for this prototype system; it allows for analog EME information as well asmodulator biasing signals to be transferred to a remote site. The optics of the sensor head ispackaged in a plastic box that mates to the wideband antenna via an SMA-connector. Antennaor probe output powers ranging from less than -60 dBm to greater than 20 dBm can be expectedat some topside positions.[12] This implies that the monitoring system must realize an SFDRexceeding 80 dB and a sensitivity below -60 dBm. This system performance must eventually beattained from 2 MHz to 50 GHz by some combination of field detection units. This will beaccomplished by combining the broadband RF information from the banded antennas onto asingle fiber optic link, as is shown in figure 2-3. Hence, optical modulators with the potentialfor extremely wide passband operation are required for this shipboard application. The opticalwaveguide modulator alternatives being pursued in this work will now be discussed.

4

PMF ( I lM).oaOTVLASER

MULTIOCTAVEANTENNA

Figure 2-1. Schematic diagram of an antenna-coupled externallymodulated EQ EME monitoring system.

PHOTOVOLTAIC

MODULATOC

OPTICAL FIBER

Figure 2-2. Optical powering scheme using GaAs photovoltaic cellsfor the E O EME monitoring system.

5

2-500-MHz O.5-18-GHz 18-50-GHzANTENNA ANTENNA ANTENNA

Figure 2-3. Triplexing scheme to combine RF information from 2 MHz through

50 GHz.

2.1.1 Lithium Niobate Optical Waveguide Modulators

The most used crystal for producing broadband optical waveguide modulators is lithiumniobate. This material has a large linear EO coefficient and low-loss single-mode waveguidescan be easily formed by using titanium indiffusion or proton-exchange techniques. MZ wave-guide modulators fabricated with traveling-wave electrodes on x-cut lithium niobate are the mostpromising of the commercially available high-bandwidth modulator devices. A schematic ofthis interferometric waveguide modulator is shown in figure 2-4. In these devices, light iscoupled from a single-mode optical fiber into a lithium niobate waveguide. The light is equallysplit at the input Y-branch and then recombined at a similia output Y-branch. The light in onearm of the interferometer is phase-modulated by an RF signal and results in intensity modulationfor the combined output optical waveguide mode. The intensity-modulated output opticalpower is then coupled back into a single-mode optical fiber for transmission. The expression forthe output optical power as a function of the input optical power for an ideal MZ modulator isgiven by

Sout(V) = Sin tm cos2 (xV/2Vx) (2.1)

whereSi = input optical powertm = modulator transmission loss factorV = modulator half-wave voltage

6

sPLITTrNG RECOMBINATION

Figure 2-4. Schematic of lithium niobate MZ interferometric optical i dItor.

For analog applications, in addition to the RF signal, a DC bias of V,/2 is applied to achievemaximum RF sensitivity. The highest speed commercially available lithium niobate MZmodulators possess an 18-GHz modulation bandwidth, a V. of approximately 10 volts, and a tmof approximately 0.5.* An x-cut lithium niobate MZ modulator with performance speciicationsclose to this has been procured and tested in this effort. The x-cut crystal was chosen over thez-cut crystal because of its demonstrated superior thermal stability. This device was used in the2- to 18-GHz laboratory system that was tested and whose experimental results will be discussedin Sections 2.3 and 2.4. Alternative optical waveguide modulators that have attracted consider-able recent attention are II-V semiconductor-based interferometric and EA modulators. Theseare now discussed.

2.1.2 rn-v Semiconductor Optical Waveguide ModulatorsGaAs-based MZ modulators operating at 1.3 tiun are beginning to become available. The

operation of these modulators is identical to that of the lithium niobate modulators discussedabove. Better phase matching between the optical and electrical signals is expected with theGaAs traveling-wave modulator than with the lithium niobate modulator. This should allow forhigher bandwidth devices without sacrificing modulation efficiency. Obtaining low opticalinsertion loss is a remaining challenge for the GaAs MZ modulators. A DC to 50-GHz proto-type GaAs modulator possessing an RF Vx = 10 V and a 10-dB optical insertion loss is to bedelivered to the Navy for use in this effort. For the bandwidth, the modulation efficiency of thisdevice exceeds any available lithium niobate modulator, although the optical insertion loss issignificantly higher (10 dB compared to 3 dB). When this semiconductor modulator is received,it will be tested and evaluated for use in this shipboard application. Another promising Ill-Vsemiconductor optical modulator is the EA waveguide modulator. This type of modulator willnow be discussed.

Semiconductor EA waveguide modulators fabricated in either a p-i-n or p-n junctionstructure are also being considered for this analog fiber optic application. Here, a reverse bias

*Available from GEC Advanced Optical Products.

7

across the junction modulates the electric field in the waveguide and changes the absorptincoefficient of the material. The semicductor EA waveguide modulator possesses an exponen-tial optical power transmission relation given by

Sow(V) = S ntm exp[ - FAa(V) L] (2.2)

whereSmn - input optical powern = modulator ansmission loss factor

L = waveguide lengthF = optical mode and active absorbing layer overlap integralzia = change in absorption coefficient

In quantum well structures, Aa is due to the QCSE. In thick (> 500 A) or bulk semiconductorlayers, Aa is due to the FKE. QCSE modulators typically have a largeeda at low appliedvoltage, but possess a relatively small F, while FKE modulators typically have a small Aa, but alarge F. It can be seen from Eq. 2.2 that these two parameters equally affect the modulator'sperformance and therefore both devices can obtain comparable modulation performance.Prebiasing the EA modulator to the quasilinear region enables analog operation in the samemanner as with MZ modulators. The modulator linearity can be assessed by measuring theharmonic and intermodulation distortion about a given bias position, which ultimately deter-mines the SFDR.

Two modulators, a 1.52-1mn QCSE modulator and a 1.32-pm FKE modulator, have beenfabricated and tested for use in the EO EME monitoring system. The liquid phase epitaxy (LPE)grown 1.32-pun FKE modulator is schematically shown in figure 2-5 and uses an InGaAsPactive absorbing waveguide layer. This ridge-waveguide modulator has a device length of 300fpm, a waveguide thickness of 0.4 pm, a device capacitance of 0.2 pF, and a r x- 0.7. For digitalapplications, an extinction ratio of > 30 dB at less than 10 V has been obtained with thismodulator.

The 1.52-pum QCSE modulator uses an InGaAs/InP quantum well ridge-waveguide structureand was grown using metalorganic chemical vapor deposition (MOCVD). A schematic of thismodulator is shown in figure 2-6. This device has a length of 650 pm, an undoped superlatticewaveguide thickness of 1 pm, and a r = 0.1. It is the ten 70-A InGaAs quantum wells buried inthe center of the superlattice waveguide of this modulator structure that are responsible for the1.52-pm absorption modulation. This waveguide modulator has also displayed a > 30-dBextinction ratio at less than 10 V applied bias. For more details about these specific modulators,the interested reader is referred elsewhere.[13.141 Analog photonic links and electromagneticfield monitoring systems employing these optical modulator types will be analyzed next.

8

AW~n P4:WNACT

- AePLATID AS lE P 4WAVEGUWDE MOVE 4

Figure 2-5. Structure of LPE-grown 1.32-Iun InGaAsPFKE optical modulator.

n. - InPSUBSTRAT CONTACT

Fiur 2-.SrctrLfMOV-rw 1.52m Ism p -sP/lODmultiple0 quan-- well QCSE opj a noduator

100 S In9

2.2 ANALYSIS OF EO EME MONITORING SYSTEM

The externally modulated EO EME system described in Section 2.1 will now be theoreticallyanalyzed. Performance will be predicted and important noise sources identified. Particularattention will be paid to the performance characteristics of the optical modulator. For digitalapplications, a modulator's usefulness is determined primarily by its bandwidth, switchingvoltage (bias required to obtain a 10-dB extinction ratio), and optical insertion loss. For analoglink applications like the one at hand, the modulator's bandwidth, linearity, RF efficiency, andoptical insertion loss are important. These parameters will now be discussed for both MZmodulator and EA modulator-based systems, and a performance comparison analysis will bemade between the lithium niobate and semiconductor modulator options.

To assess the relative usefulness of the candidate optical waveguide modulators discussed inSection 2.1, an externally modulated fiber optic link was breadboarded. Modulator characteris-tics were measured for the FKE and QCSE modulators and compared to those of a high-speed1.32-nIm lithium niobate MZ waveguide modulator. The MZ modulator performance specifica-tions used for this comparison are a V, = 10 V, a fiber-to-fiber insertion loss of 3 dB, and a 3-dBelectrical bandwidth of 18 GHz, specifications which at this time compare favorably with othercommercially available MZ modulators. As described in the previous Section, the releventQCSE modulator parameters for transverse magnetic (TM) mode operation are a device lengthof L = 650 gjm, a confinement factor of F = 0.07, and a device capacitance of 1.2 pF. Therelevent FKE modulator parameters are a device length of 300 Iim, a confinement factor of 0.7,and a device capacitance of 0.2 pF.

The relative transmission versus applied bias or optical transfer functions for the FKE,QCSE, and MZ modulators are shown in figure 2-7. The important aspects of these curves forthis analog application are (1) the bias required to obtain quasilinear operation, (2) the slope ofthe curve at the bias point, which is a measure of the RF efficiency, and (3) the linearity aboutthe bias point. These aspects will now be investigated.

The externally modulated fiber optic link RF insertion loss can be expressed asH 2' K' r2 oR r2 (2.3)

f d Out

whereKf = optical loss from source to detector (not including modulator insertion loss)rd = detector responsivity (AIWo)Rout= detector output resistance (0)r = modulator RF efficiency factor (Wo/We -).

For an MZ modulator

r12 = (tin Po ar/VX)2 (Rm/2) (1-p2) (2.4)

where P0 is the laser power, tm is the modulator transmission factor, Rm is the modulator input resis-tance, and pm is the modulator input RF reflection coefficient. For an EA modulator, rm is givenby

r2 = (tin Po)2 (Rm/2) (1-p 2) [/L (dAa/dV)vb]2 (2.5)

10

1.0

0.8

PAZ

0.4

n-

0.2OCSE

0 .0 -

0.0 2.5 5.0 7.5 10.0

APPUED BIAS (V)

Figure 2-7. Relative transmission versus applied bias curves for the1.32--tm FKE waveguide modulator, the 1.52-!.n QCSE waveguidemodulator, and the 1.32-pm lithium niobate MZ waveguide modulator.

where the derivative is taken at the modulator bias voltage, Vb. Measured values for A a as afunction of voltage for the QCSE and FKE modulators are shown in figure 2-8. It is found that(da/dV) evaluated at Vb - 2 V is equal to 150 (cm-V)-1 for the QCSE modulator and 50(cm-V)"' at Vb - 6 V for the FKE modulator. If we assume that P, = 50 mW, Rm = 50 Q, P,=0, and tm = 0.10, we find that rFKE2 = 6.9 x 10-4 (Wo2/We)and rOCSE2 = 2.9 x 10-4 (Wo2/We). Forthe MZ modulator, we will also assume Po = 50 mW, Rm = 50 Q Pm= 0, but a lower opticalinsertion loss yielding tm = 0.50. Using these numbers gives rMz 2 = 1.5 x 10.3 (Wo2/We) whichis slightly larger than that of the FKE modulator. This result is entirely due to the presentlyencountered larger EA modulators' optical insertion loss (10 dB versus 3 dB). Hence, eventhough the insertion loss of the EA modulators is substantially larger than that of the MZmodulator, the RF power efficiencies are comparable due to the larger modulator optical powertransfer function slopes evident in figure 2-7. If the insertion loss of the EA modulators can beimproved to the same level as the MZ modulator, then efficiencies of rFKE2 = 1.7 x 10-2

(Wo2/We) and rOCSE2 = 7.3 x 10-3 (Wo2/We) are obtained, which are significant improvementsover current state-of-the-art MZ modulators. If we further assume that Ro.t = 50 Q, rd = 0.85A/IW, and Kf = 0.1 (including 3 dB for biasing of the modulator), then the expected RF insertionloss is -33 dB for the MZ system, -22 dB for the FKE modulator system, and -26 dB for theQCSE modulator system. These results translate into a lower system noise figure (NF) for theEA modulators if low optical insertion loss devices can be developed. It is important to note thatthe QCSE modulator structure used in these experiments is not optimized. This optimization

11

process involves improving the material quality of the quantum wells as well as modifying themodulator geometry to increase r. When optimized, the performance of the QCSE modulatorshould be comparable to that demonstrated by the FKE modulator.

2MOO

OCSE

-1500

0E0

LU

500

FKE

0~

0 2 4 6 8 10APPUED VOLTAGE (V)

Figure 2-8. Experimental values for Aa versus applied bias for the1.32- jm FKE and 1.52-Rm QCSE waveguide modulators.

The SFDR of the MZ, FKE, and QCSE modulator systems will now be addressed. For theexternally modulated link, with the above-used laser, detector, and laser-to-detector lossparameters, a detector optical power of 2.5 mW is expected for each link. Again, this assumes4,, = 0.5 for all three systems. With these detector optical power levels, as will be discussed inSection 2.2.2, the fiber optic links are shot-noise-limited. This assumes a laser relative intensitynoise (RIN) of -165 dBc/Hz, which is well below the thermal and shot-noise contributions atthese detector optical powers. This demonstrates the importance of minimizing the modulatorinsertion loss as well as using a high-power, low RIN laser in obtaining shot-noise-limitedoperation.

The receiver signal power and spurious signals due to intermodulation products and harmon-ics for the MZ, FKE, and OCSE modulator systems are graphed versus modulation index infigures 2-9 to 2-11, respectively. A normalized receiver bandwidth of 1 Hz, which determinesthe system noise floor, is used for these simulations. The result is that a 108-dB SFDR in a 1-Hzbandwidth can be expected for the MZ system. This result agrees well with actual two-tonedistortion measurements performed on this link which yielded a 102-dB/Hz 2/3 SFDR with a10-mW laser power and a 109-dB/Hz2/ 3 SFDR with a 50-mW laser. Ideal biasing at thequadrature point is assumed for this calculation. The spurious signals are obtained by expanding

12

-4064

S-80 B

10- SI•NAL100i~-140 14

-1604-

-180:- NOISE FO / .MoNI 18O

L0OULATION INDEX

Figure 2-9. Optical receiver output power versus modulation index, includingharmonics and intermodulation products for the 1.32-jum MZ modulator.

102dB,.. I

S-90HNI INTERMODVD

2: - 100- SINL-120:62

o 143HARhIONC-4. 160

NOISE FLOOR tB

-20 ....LI.ZV /"-L I .. LAJL."" .~LUi2Z0 M0 0 CV 0

WI W W W

MODULATION INDEX

Figure 2-10. Optical receiver output power versus modulation index, including

harmonics and intermodulation products for the 1.32-jun FKE modulator.

13

- 20E

I ] I I I I I I

IJ LW Li Li .i Li L

103OIIATION INDEX

Figure 2-11 . Optical receiver output power versus modulation index, includingharmonica and intermodulation products for the 1.5210m QCSE modulator.

the modulator optical power transfer function in a Taylor series and evaluating the nonlinearterms appropriately.[15] Using the experimental modulator transfer curves of figure 2-7, SFDRsof 102 dB and 105 dB in 1-Hz bandwidths are obtained for the FKE and QCSE modulatorsystems, respectively. Hence, not much difference in dynamic range performance is expectedusing either an BA or MZ optimally biased modulator. At the chosen bias positions, all three ofthese modulators were limited by the third-order intermodulation product and not the fundamen-

tal harmonics. For broadband systems such as in shipboard EME monitoring, the harmonics aswell as the intermodulation products are important It is important to note that the resultantdynamic ranges for the BA modulators were not as sensitive to bias position as the MZ modula-tor dynamic range within the 0.2 to 0.8 relative transmission ranges. fiis indicates that theSFDR performance of the exponential transfer function is less sensitive to bias position than thecosine-squared transfer function. This can be important in field-deployed situations whereenvironmental fluctuations, which can affect the modulator bias position, are unavoidable. Inthe case of temperature fluctuations, significant modulator bias point drift can be expected.Hence, the FA modulators seem to be very competitive with the state-of-the-art lithium niobateMZ modulator in link performance.

The electrical power out of the antenna can be expressed in terms of the externally modu-lated fiber optic link parameters as

P = (tin P0 m/2rn)2 (2.6)where f is the modulation index. For the MZ system, this gives

PA= (in2 V=)/[2g2 Rm (1 - p2](2.7)

14

and for the EA system

= m2/[2 Rm (I - pI) P L 2 adV)2)](2.8)

Using the minimum detectable modulation indices given in figures 2-9 to 2-11, the minimumdetectable antenna power in a 1-Hz receiver bandwidth is PA (min) = -135 dBm for the MZsystem, PA(min) = -146 dBm for the FKE system, and PA(min) = -142 dBm for the QCSEsystem. Again, these numbers assume equal modulator optical insertion losses of 3 dB, anassumption which remains to be demonstrated for the semiconductor devices. Nevertheless,excellent link sensitivity is expected, which translates into fairly low electric field sensitivities.

The system bandwidth of the EO EME monitoring system is ultimately limited by themodulation bandwidth of the optical modulators. For the MZ modulator, the bandwidth is 18GHz with an assumed V. = 10 V. A high-frequency V, of closer to 30 V is actually measuredwith this lithium niobate MZ modulator. If the manufacturer-specified V, value is used, a0.56 V/GHz bandwidth-efficiency figure of merit results, which when extrapolated to beyond 40GHz, implies devices with V. - 25 V. For the EA modulators, device capacitances in the0.2-pF range are routinely attained with the FKE devices, which implies cutoff frequenciesgreater than 30 GHz with Vb = 5 V. An experimental frequency response curve of the the1.32-$tm FKE modulator is shown in figure 2-12. At 20 GHz, the highest measurementfrequency currently available, the response is less than 1 dBe down from its low-frequencyresponse. This device possessed a 7-V on/off voltage, which translates into a bandwidth-effi-ciency figure of merit of better than 0.5-V/GHz, which already exceeds that of current lithiumniobate MZ modulators. With optimized EA modulator structures, it is expected that the biasvoltages required to obtain these extremely large bandwidths can be reduced to less than 5 V,which is a significant improvement over that attainable with comparable lithium niobate MZmodulators. A 1.5-pmo quantum well waveguide modulator with a 3-dBe bandwidth exceeding40 GHz at 5-V bias voltage has been reported,[ 161 which does demonstrate the usefulness ofthese modulators for ultrawideband applications.

The most useful and revealing parameters to evaluate the performance of the fiberoptic sensing system are the link signal to-noise (S/N) ratio and the link NF. The S/N ratio willbe considered first. The available output electrical signal power is given by

Psig = (m2/2) 12 Rout (2.9)

where ID is the optical detector time averaged photocurrent, and the other parameters werepreviously defined. Three system noise sources are considered important: shot noise, thermalnoise, and laser RIN noise. The available shot noise power is given by

Psho = 2 q ID Row B (2.10)

where q is the electronic charge and B is the receiver bandwidth. The available thermal noisepower is given by

Pth = K T B (2.11)

15

>521 ~~log MA13KME -REF -70.0 dB W/A 6.6 Hz

S3.6 d8/ 6.6 dO-0. 9023 dO

F'rmi'nic" Modulator -L - .-- is 187 H

SCA-E -6.9623 dBs 3.0 IB/€ iv

H _

A

S6 Jan 1994 15:45START 6. 136660666 0H-zSTOP 26.666666666 0Hz

Figure 2-12. Frequency response curve for 1 .32-tpn LPE-grownFKE modulator.

where K is Boltzman's zonstant and T is the receiver temperature. At room temperature, Pth =

-174 dBm/Hz. The available laser noise power is given by

PL = 12 RIN Ro, B (2.12)

where the RIN value is usually specified for each laser or can be measured. The inverse of the linkS/N ratio can then be expressed as

1/(S/N) = 2B/r 2 [(RIN) + (2 q/0D) + (KT/RoPJID)] (2.13)

where the first term in the bracket represents the laser noise contribution, the second term representsthe shot noise contribution, and the third term represents the thermal noise contribution. Dependingon the values of RIN, Rout, and ID, the system is either shot-noise-limited, RIN-limited, or thermal-noise-limited. A plot of m21(S/N) in a 1-Hz bandwidth versus ID is shown in figure 2-13 for Rout= 50 Q and different values of the laser RIN. This curve demonstrates the importance of obtaininga laser with a very low RIN value.

The fiber optic link NF is probably the most important parameter which can be measured, for

it ultimately determines the detection sensitivity limit. The link NF is defined as

NP = 10 loglo [(S/N)jnpu(S/N)outPUt] (2.14)

which can be expressed in terms of the link parameters as

NF - - 10 log o(H2) + 10 log10 [Pth(l + H2) + PL + Pshaot]IPh (2.15)

16

-110 . .

-120:

-130'-

-140S"• RIN =-1:30 dBc/Hz

-150

CI

-160- -145 dBc/Hz-

-170-

-180-173 dBc/Hz

-190-

10-6 i0-5 i0-4 10-3 10-2 10-1 100

DETECTOR PHOTOCURRENT (A)

Figure 2-13. Plot of m2/(SIN), where m is the modulation index, in a 1-Hzbandwidth versus optical detector photocurrent for different laser RIN values.

where H2 is the RF insertion loss given by Eq. 2.3. It is assumed in this NF expression that theinput and output thermal noise powers are identical. For the MZ and EA modulator links, theexpected NFs are 28 dB for the FKE system, 32 dB for the QCSE system, and 39 dB for the MZsystem. Again, a clear advantage for the EA modulators if low optical insertion loss can b,achieved. Experimental results from a breadboarded MZ fiber optic link will now be discussedand compared with these simulated results.

2.3 PERFORMANCE OF FIBER OPTIC LINK

In this section, the best available 2-MHz to 18-GHz fiber optic link will be described. Adiagram of the externally modulated fiber optic link is shown in figure 2-14. The 1.32-pimNd:YAG laser (Amoco Model ALC1320-56S) has a fiber pigtail output power of 45 mW and ameasured RIN value of approximately -170 dBc/Hz. The lithium niobate MZ modulator(GEC-Marconi Model Y-35-8808) has an optical insertion loss of 3.6 dB and a low frequencyV, = 13.5 V. The modulation frequency response of this modulator is shown in figure 2-15,which reveals a 3-dBe bandwidth of approximately 17 GHz. The InGaAs p-i-n photodiode(Fermionics Model HSD-30) has a fiber-coupled responsivity of rd = 0.7 A/W at an opticalwavelength of 1.3 fVm. A 3-dBe modulation bandwidth exceeding 16 GHz was measured forthis optical detector. The overall fiber link frequency response is plotted in figure 2-16, whichshows a 10-dBe falloff in response from 50 MHz to 18 GHz (3-dBe bandwidth of 11 GHz).The fiber link loss budget is depicted in figure 2-17, which displays an optical loss from laserto detector of 10 dB, a value that includes the 3 dB for modulator biasing. The total noisepower (Pth + Pshot +PL) of this link as a function of optical detector photocurrent is plotted in

17

45-MW SINGLE-MODE1.32-pm Nd:YAG LASER

15-GHzLITHIUM NIOBATE

U Z MODULATOR

RFRF INPUT

SMF (300 m)

18-GHzInGaAs p-i-nPHOTODIODE

Figure 2-14. Schematic diagram of best available DC to 18-GHzexternally modulated fiber optic link.

REF -29.98 dDW/A U.0. HzS3.0 dlW 0.0 do-2.916 di s

> )MARKER 2-11SC, '17 .0 5 2 G 1"1 'A 'SCALE : ;,-2.916 dD

s 3.0 NB/div v _ l '.

HI .. I

S7RT 0.SSSSS .:10 Fob 1994 15:33

STOP 19.000000000G mi

Figure 2-15. Modulation frequency response of wideband lithiumniobate optical modulator.

18

>s21 log MnA nwm]l-i=•1RF-40-0 01B 0.0. HaS3. 0 411/ 9.0 do-7.2998 dB4

| _!_____�_______IM )W4ER 2-1'r-C--------- 16.065 -Hz:AISCALII I Z 1-7.2999 dl,S I 3.0 •IB/4

IAH .-- -- - . . t- --- i ° ---- • '

SH

H-- rI - - t' -- I •IN,

*I I I l _

START 0. 656666660 0HzDIP~ 18.000000090 0Hz

-Figure 2-16. Externally modulated fiber optic link RF insertion lossversus frequency in the 50-MHz to 18-GHz range.

(45 mW) (4.5 mW)

~ * ~ 0EiCTOR.

_ -.. dB -1 dB'tAS I I II

-7 dBFigure 2-17. Diagram of the externally modulated fiber optic link loss budget.

19

-150 .-155- DATA OPERATING I /

-155 ~POINT j/-160-

-- - 7-THERMALR-175-

Z SHOT /a- /

-8/ RIN --173 dBc/Hz-a-190

I II I I IIl10-6 10)-5 10-4 10-3 10-2 10-1

DETECTOR PHOTOCURRENT (A)

Figure 2-18. Total noise power (solid line) and experimental data (circles) versusoptical detector photocurrent for the externally modulated fiber optic link.

figure 2-18, which shows that the operating point for this system renders shot-noise-limited opera-tion. The RF insertion loss of this link was measured to be 32 dB at 2 GHz. This is within 3 dB ofthe theoretical prediction using Eq. 2.3. The link NF was measured to be 35 dB, 38 dB, and 45 dBat frequencies of 2, 10, and 18 GHz, respectively. The insertion loss and NF values can be improvedupon by increasing the laser power and decreasing the optical modulator V= voltage. For compari-son, the RF insertion loss versus frequency of a 2-MHz to 500-Mi-z fiber optic link using a opticalmodulator possessing a V. of 4 V is shown in figure 2-19. Here, a link RF insertion loss of 11 dBand an NF of less than 20 dB are obtained, which demonstrates the importance of maximizing themodulation responsivity of the optical modulator. With continued laser and ultrawideband modula-tor R&D, the potential for fiber links with low RF insertion loss can be expected. Two-tone measure-ments with the 18-GHz fiber optic link show an SFDR of 108 dB/Hz2/3. The SFDR is smaller thanthe linear dynamic range for this wideband fiber optic system which implies a linear dynamic rangegreater than 110 dB/IIz. This link will be field-tested and subsequent shipboard measurements willbe acquired.

2.4 ELECTROMAGNETIC FIELD MEASUREMENTS IN THE 2- to 18-GHZFREQUENCY RANGE USING AN EO EME MONITORING SYSTEM

Remote electromagnetic field measurements in the 2- to 18-GHz frequency range using anantenna-coupled externally modulated fiber optic link will be described in this Section. Thesemeasurements have been made with a fiber optic link whose performance was not quite as goodas that described in the previous Section. Nevertheless, an electric field sensitivity of 15 WtV/mand an SFDR of 102 dB in a 1-Hz bandwidth have been measured. The results indicate that thissystem is a feasible candidate for remotely measuring extremely broadband, large dynamic rangeelectromagnetic fields, especially when recent fiber optic link improvements are incorporated.

20

)Oal leg W

1. -. a i

Figure 2-19. RF frequency for a DC to

500-MHz fiber optic link using a V/ = 4 V optical modulator.

2.4.1 System Configuration

A schematic diagram of the antenna-coupled electromagnetic (EM) field detection systemused in this early work is shown in figure 2-20. It consists of a cavity-backed spiral antenna(Transco Model 9C33500), a Ti:indiffused lithium niobate MZ waveguide modulator (GEC-Marconi Model Y-35-8808), a 1.32-tim Nd:YAG solid-state laser (Amoco ModelALC1320-10S), a 100-meter length of PMF single-mode optical fiber (Alcoa-Fujikura ModelN000185), a 100-meter length of SMF (Coming Model SMF28), and a high-speed InGaAs p-i-nphotodiode (BT&.D Model PDC4310-30-FP). The electrical output of the broadband antennawith known characteristics is coupled to the optical modulator and is optically remoted via PMFfor the upiink and SMF for the downlink, with the laser and optical detector remotely locatedwith the associated processing electronics. In an effort to limit the electrical power requirementsand intrusiveness of the EM field probe, this configuration contains no low-noise amplifier at theinput to the optical modulator. The externally modulated fiber optic link is slightly differentthan that described in Section 2.3. The Nd:YAG laser has an output power of 10 mW and ameasured RIN of -173 dBc/Hz. The lithium niobate MZ modulator has a 3-dBe modulationbandwidth of 18 GHz, a 3.6-dB optical insertion loss, and a low-frequency half-wave voltage ofVx = 13.5 V. The optical detector has a responsivity of rd = 0.85 A/W and a 3-dBe modulationbandwidth of 25 GHz. The fiber optic link and antenna performance were first measured andcalibrated separately and then mated for anechoic chamber testing of the entire EM fielddetection system. The results of these experiments will now be discussed.

21

I~TQPOJ~ PUF (100 m)SMF~t (100-.NM)

Figure 2-20. Schematic diagram of the antenna-coupled externallymodulated fiber optic link for EM field detection.

2.4.2 Results

The fiber optic link and spiral antenna individual system responses in the 2- to 18-GHzfrequency range are shown in figure 2-21. The close to 6-dBe roll off in response for the fiberoptic link is due to the combined response of the integrated optical modulator (IOM) and theoptical detector. The optical insertion loss of this link is 9.0 dB, which includes the 3 dB loss forIOM biasing. The RF insertion loss of this link was measured to be 45 dB at a frequency of2 GHz. A lower RF insertion loss and hence better performance has since been obtained byemploying a higher power laser. The effect of laser optical power on fiber optic link perfor-mance is better illustrated in figure 2-22, which graphs the inverse S/N ratio as a function ofdetector photocurrent. The individual thermal, shot, and RIN noise contributions are plotted toillustrate the thermal-noise-limited, shot-noise-limited, and RIN-limited regions of operation,The operating point for this fiber optic link is borderline between thermal-noise-limited andshot-noise-limited. More optical power would increase the SIN ratio and decrease the NF of thelink. Link NF values of 49 dB, 52 dB, and 55 dB have been measured at modulation frequenciesof 2 GHz, 10 GHz, and 18 GHz, respectively. Two-tone distortion me ents were per-formed near 9.5 GHz to determine the SFDR of the link. The theoretical and experimentalresults of these measurements are summarized in figure 2-23. A 102-dB/Hz2 SFDR has beenempirically measured by fitting the experimental data points for the fundamental and third-orderintermodulation power levels to theoretical curves. Excellent agreement is obtained when anIOM Vx of 30 V is assumed instead of the 13.5-V low-frequency value. This discrepancy is notthat alarming considering that the two-tone measurements were performed at 9.5 GHz and theVx of 13.5 V was essentially measured at DC.

22

- .40n dulo~flM *** ll',% I* -" - -ý Z-

SII,

1 11.01 • 0. a1, do

STAWVr 2. 9Wz 22 Now 181l l0S1STO 19.0000o1, o 9 4z

(a) Fiber optic link.

4--

2-6---- e0-"

-2 VS ?R: 2.:

-4"POIARITIONILHC

"BEIMWtTH (3dB): D DE

-101-

2 4 6 8 10 12 14 16 18FREQUENCY (GHz)

(b) Spiral antenna.

Figure 2-2 1. Modulation frequency response from 2 to18 GHz of the externally modulated fiber optic linkand the spiral antenna.

23

-110

a DATA-120

THERMAL NOISE LMT

-130

-140

1 "5- OPERATING

SHOT NOISE LII-160-

-170-

-180

-190 RIN LIMIT

10-6 105 10-4 10-3 10-2 10-1 100

DET'ECOR PHOTOCURRENT (A)

Figur 2-22. Plot of m21(S/N), where m is the modulation index, fortotal noise (solid line), shot noise, thermal noise, and RIN contributionsversus detector photocurrent for the fiber optic link used in this work.

"-20- MODULATOR VX z 30 VRIF INSERTION LOSS a, 48.9 dB v .;\,

-40 THIRD ORDER INTERCEPT 3 8.1 dBmSFDR a95.3 dB

-60 NOISE BANDWIDTH= 1 Hz

-go-

-100-1

Ef

-160 NOISE FLOOR

-120 -100 -80 -60 -40 -20 0 20 40

MODULATOR DRIVE POWER (d0l)

Figure 2-23. Plot of experimental (circles) and theoretical detectedpower for fundamental (solid line) and third-order intermodulationproducts (broken line) versus power applied to the IOM.

24

The wideband RF probe used for this work is a 2-inch-diameter cavity-backed spiral antennawhose gain characterisitics were given in Figure 2-2 1b. It is a left-hand circularly polarizedantenna that has a 3-dB beamwidth of 80 degrees and a voltage standing wave ratio (VSWR) ofless than 2:1 across the 2- to 18-GHz frequency range. The gain versus frequency characteristicswere measured against standard gain horn antennas in the 2- to 18-GHz range for linearlypolarized radiation, and > 0 dBi gain exists for frequencies above 4 GHz. The circularlypolarized spiral antenna is a good choice for multioctave frequency coverage if a small, passivereceive antenna is required.

EM field measurements were made with the antenna-coupled fiber optic link after separatetesting of the antenna and link was completed. A 2- to 18-GHz far-field anechoic chamber wasused in which to perform the experiments. Measurements were performed at 2.25, 9.52, and 16GHz using three different standard gain horn source antennas. The spiral antenna outputelectrical power was calculated from the expression

PA = Aef Winc

= (A2GA/4,ir) Winc (2.16)

= (PM,) (Gtr) (GA) ( 2/4xR)2

where

Aeff effective aperture of spiral antenna (m2)

inc f RF intensity incident on antenna (W/m 2)

Pin = RF power applied to transmit antenna (W)

Gtr = gain of transmit antenna (dBi)

GA = gain of spiral antenna (dBi)

X. = RF source wavelength (m)R = propagation length (m)

The "actual" electric field strength at the probe site was calculated from the expression

EA = (2 t/ Wnc) 1/ 2 (V/m) (2.16)

where ij = 377 Q is the free space impedance. The "measured" electric field strength wasobtained using Eq. 2.17 along with the fiber optic link RF insertion loss data. Starting with theelectrical power out of the optical detector, PA was deduced by factoring in the fiber optic linkloss. From PA, the incident RF intensity WV11 was determined, and from that the electric fieldstrength was found. It is important to have an accurate calibration of both the spiral antennagain and the fiber optic link loss as a function of frequency if precise electric field measurementsare to be made.

The EM field detection system response for boresight radiation at the three frequenciesinvestigated is shown in figure 2-24. In this figure, the electrical output power of the optical

25

detector is plotted versus incident RMS electric field strength. A 10-Hz spectrum analyzerresolution bandwidth was used in this measurement. Extremely linear response is obtained fromthe highest field levels down to the minimum detectable levels for each frequency investigated.To assess the accuracy of these field measurements, electric field values were obtained from thedetected optical power levels, since the RF insertion loss of the link as well as the gain of theantenna were known. The result of this measurement is shown in Figure 2-25 where measuredfield strengths are plotted versus calculated electric field strengths. There is some deviationfrom expected field levels, which is attributed to the anechoic chamber testing procedure. Thepointing accuracy for the three relatively high gain horn antennas was not precisely controlled,resulting in actual electric field levels that differed from the calculated levels. Nonetheless,RMS electric field sensitivities of approximately 15 IV/m, 44 VV/m, and 107 •IV/m have beenachieved in a 1-Hz resolution bandwidth for the 2.25, 9.52, and 16-GHz frequencies, respec-tively. The RMS electric field detection ranges for the three frequencies investigated aresummarized in table 2.1. These experimentally attained electric field detection ranges imply thatthis system can be useful for remotely performing broadband, large dynamic range electric fieldmeasurements.

-6O

4 2.25 GHz 4+

* 9.52 GHz 4+-0 a 16.0GHz .4.

N• 4. oO -100 +.

44 o4+

S-120 +

o 4. .I�14 + 13

o 0 ql4r 4.Oa. +1

0 +

-160 • o4.D+W • • O•

-18 ... ! . .. " .. - .. • . ..- ' .. ......

-180

10"6 10-5 10- 10-3 10-2 10-1 100 101

CALCULATED ELECTRIC FIELD (V/m)

Figure 2-24. EM field detection system output versus incidentRMS electric field level.

26

101101 - THEORY +

+2.25 GHZ +j

100 9.52GHz + +

a 18.OGz U

S10"1 413

10-2•4p

a .

10 ..4÷10-3

-', + +

-5

10

10-6 10 -4 10-3 10-2 10-1 10 0 10 1

CALCULATED ELECTRIC FIELD (V/m)

Figur' 2-z5. Measured RMS electric field levels versuscalculated RMS electric field levels.

Table 2.1. Externally modulated EO EME monitoring systemelectric field level detection ranges at 2.25, 9.52, and 16.0 GHz.

RMS Electric Field Detection Range

Frequency (1 Hz Bandwidth) (1 kHz Bandwidth)(GHz)

2.25 15 t&V/m - 0.8 V/m 0.02 V/m - 90 V/m9.52 44 •IV/m - 6 V/m 0.04 V/m - 550 V/m

16.0 107 IAV/m- 17 V/m 0.1 V/m - 1585 V/m

27

3.0 IMPROVEMENTS TO EO EME MONITORING SYSTEM

The Phase I testing of the breadboarded EO EME monitoring systems proved quite valuablein assessing the technology and determining areas of further development. In this Section,system improvements and modifications which have been implemented as part of the Phase H1efforts associated with this program will be discussed. Performance improvements in the fiberoptic link and wideband RF probes have been attained. One improvement is the development ofstable, remote optically powered and controlled fiber optic links, and another is remote polariza-tion control of fiber optic links. Link performance improvements have been achieved due to theinsertion of superior optical components. The current status of the ultrawideband antennadevelopment efforts will be reviewed. The fiber optic link improvements and modifications willbe discussed first.

3.1 IMPROVEMENTS TO FIBER OPTIC LINK

The performance of analog photonic components has been steadily improving year by year.Since the start of this R&D effort in FY 91, significant advances have been made in the areas ofsolid-state laser and optical modulator technology, and these advances affect the performance ofthe EQ EME monitoring system. In this section, fiber optic analog link performance improve-ments and link modifications during Phase II of this effort will be discussed.

The link performance of the "best available" 18-GHz fiber optic link has been measured.This link is similar to that described in Section 2.3, but has slightly superior optical componentsand is the link to be used in the FY 94 shipboard demonstration under this project. It consists ofa 50-mW Nd:YAG laser (Amoco Laser Company Model ALC1320-50S), a 300-meter length ofPMF (3M Model FS-HB-6621) for the uplink, a 300-meter length of SMF (Corning ModelSMF28) for the downlink, an 18-GHz optical modulator (UTP Model APE MZM-1.3-18-T-01),and an 18-GHz InGaAs p-i-n photodiode (Fermionics Corporation Model HSD30). Postdetec-tion amplification with a low-noise amplifier (LNA) is implemented to reduce the overall NF ofthe receiver. The spectrum analyzer used in this work (Hewlett Packard Model 8566B) has anNF of approximately 40 dB.

The fiber optic link transfer function (electrical in/electrical out) is shown in figure 3-1. AnRF insertion loss of 35 dB is shown at 18 GHz along with a 3-dBe link bandwidth of 15 GHz.This system possesses an NF of less than 35 dB at 2 GHz as well as an SFDR of >100 dB/Hz/.This link has been calibrated, packaged, and is currently undergoing environmental testing,which continues into Phase III of this effort.

To address the > 18-GHz system requirements, current R&D efforts include a 40-GHzlithium niobate MZ modulator being developed by Boeing, a 40-GHz MI-V semiconductor EAmodulator being developed by Fermionics, and a 50-GHz M1-V semiconductor MZ deviceprocured from GEC-Marconi Materials Technology. The status of these R&D efforts will nowbe discussed.

3.1.1 Ultrawideband Optical Modulators

Efforts to further develop the semiconductor EA modulators for shipboard analog photoniclink applications are continuing. A Navy development program with Fermionics Corporation isdirectly supporting the shipboard EME monitoring effort. The purpose of this Office of Naval

28

>S21 log MAG3REF -30.0 dD 2.0 l-Iz1 3.0 dB/ -29.397 dOV -29.397 dBhP

C-A IF AVERAGIG FACTOR

START 2.SSSOOSO 004HzSTOP 18.OOUOOS I0,.Hz

Figure 3-1. Fiber optic link transfer function.

Research-sponsored and NRaD-administered effort is to develop a high-speed M-V semiconduc-tor EA modulator. The objective of the work is to commercialize a 40-GHz InGaAsP/InP FKEmodulator operable at a wavelength of 1.32 tM for digital as well as analog optical linkapplications. The operation of the FKE modulator was discussed in Section 2.1.2 of this report.At present, a >20-GHz fiber-pigtailed modulator has been delivered to the Navy.[17.18] Thisdevice suffers from a rather large optical insertion loss (8.8 dB) and only a moderate RFefficiency (10 dB with 7 V voltage swing). Theoretical simulations and other experimental worksuggest that these parameters can be significantly improved. Nevertheless, this device is beingtested for shipboard use. Current work is focused upon improving the modulator performanceparameters and delivering a 40-GHz semiconductor EA modulator in FY 94.

Under direct NRaD sponsorship supporting the EO EME monitoring task, Boeing isdeveloping a wideband traveling-wave lithium niobate MZ modulator. The goal of this effort isto deliver an MZ modulator which is suitable for operation at 1.3 ain, is operable from DC to 40GHz, has a V. of 20 dB, and a fiber-to-fiber opticalinsertion loss of 20-GHz bandwidth optical modulatorsand it intends to slightly modify the structure to efficiently achieve a >40-GHz modulationbandwidth. A 40-GHz lithium niobate optical modulator possessing the performance specifica-tions stated above is to be delivered to the Navy for testing in FY 94.

A wideband HI-V semiconductor MZ optical modulator has also been procured fromGEC-Marconi Materials Technolgy. According to specifications, this device is to operate at 1.3Ium, be operable from DC to 50 GHz, and have an RF V. = 10 V, a >20-dB extinction ratio, andan optical insertion loss of

areas except the optical insertion loss. As this technology advances, lower fiber-to-fiber opticalinsertion losses should be attained. The GaAs-based modulator with a < 11-dB optical insertionloss is to be delivered to the Navy for testing in FY 94.

The suitability of these three wideband optical modulator approaches for shipboard EMEmonitoring will be experimentafiy determined in FY 94 and recommendations will be forthcom-ing.

3.1.2 Optically Powered and Controlled Remote Fiber Optic Links

Many fiber optic antenna remoting applications, including shipboard EME monitoring,require standoff electrical powering of the antenna-coupled components. Self-contained batterypacks are one solution. However, batteries possess a finite lifetime that can limit their usefulnessfor many applications. In some cases, the optical powering can be accomplished using a PBLsystem, which has the advantage that all-dielectric fiber optic cables can be used. This can beessential for applications like the EME monitoring system, where EMI effects are to be mini-mized. In addition, for multioctave externally modulated fiber optic links, the electrical biasingof the optical modulator is critical in minimizing nonlinear distortion. Although accuratemethods of passively biasing the modulators are being investigated,[ 19] the environmentalstability of the bias position has not been established. Consequently, active modulator biasing isalso under investigation. Combining remote optical powering and remote optical modulator biascontrol has been demonstrated.[2°,2 1] Here, we will demonstrate this technique with twomultioctave fiber links and show how accurately the bias point must be controlled in order for nodynamic range penalty to result.

The two externally modulated fiber optic links investigated in this work operated from 30 to500 MHz (proton exchange modulator) and from 2 to 18 GHz (titanium diffused modulator).Each link consists of a 1.32-pgm Nd:YAG laser, PMF for the uplink, a lithium niobate MZmodulator, SMF for the downlink, and a high-speed photodiode. The PBL system consists of ahigh-power AlGaAs laser diode array, a multimode optical fiber (100 pm/140 pgm), and ahigh-efficiency GaAs photocell. The AlGaAs laser diode is intensity-modulated at low fre-quency (= 100 Hz) to introduce a reference modulation onto the optical modulator. The secondharmonic of this low-frequency modulation is then remotely detected and analyzed using alock-in amplifier. The output of the lock-in amplifier is used to actively control the opticalpower of the AIGaAs laser diode, which in turn controls the optical modulator bias position. Aschematic of this fiber optic system is shown in figure 3-2. The performance requirements andthe ability of this optical powering and biasing control technique to minimize modulator-inducedharmonic distortion ov,_; wide temperature ranges will now be discussed.

The PBL unit consists of a 250-mW AlGaAs laser diode array (Spectra Diode Labs. ModelSDL-2432-P2) emitting at 810 nm which is directly coupled into an optical fiber with a corediameter of 100 Rm (manufactured by Spectran Corp.). At the modulator side of the link, theoptical power is converted to electrical power using a 12-V GaAs photocell (Photonic PowerSystems Model PPC-12S-ST). Approximately 50 mW of optical power is delivered through theoptical link to the photocell. A graph of the output voltage as a function of series resistance foran input power of 50 mW is shown in figure 3-3. A series resistance of 2.7 k,, is chosen whichallows for an output voltage of between 0 and 12 V to be obtained. A plot of the output voltageas a function of input optical power is shown in figure 3-4. Very little current is required(

104*OSCILLATORWVITH PC OFFSET

CPU

LOCK-ORAMPLIFIER

DC SECTUMT ESRHEDASU

ANALYZER

IIF

PHOTODI0ODE

0.11.0•m MM

100 a

OPTIAL RIO

pup2-N CANXE (100 soN4:YVAG LAWNR

I-I010a

RESITANCE(KL O-HS

r,2. Schematic of orti I powering system and modulator bias control circuitry.

143

12"

10"

4-

2-

0 1'0 ... 300' O 40RESISTANCE (KILO-OHMS)

Figure 3-3. Photocell output electrical voltage as a function of seriesresistance for 50-mW input optical power.

31

12

10

w

4

2

0 25 50 75 100 125 150 175 200 225 250OPTICAL POWER (mW)

Figure 3-4. Photocell output voltage versus input optical power using a2.7-kQ series resistance.

(voltage X current) is required for this remote unit. A 0.47-piF capacitor is placed in series withthe 2.7-kQ resistor to damp out voltage fluctuations caused by transmitted optical powerfluctuations through the multimode fiber. The transmitted optical power is modulated with a100-Hz small modulation depth signal for the modulator bias control. This optical poweringtechnique allows for the modulator voltage to easily be changed simply by varying the trans-mitted optical power.

The requirement for the fiber optic link is that the SFDR is not reduced by any error inbiasing the modulator. The reciever signal powers for the fundamental through the third-orderintermodulation product for the MZ link excited with equal intensity RF tones at w, and w2 aregiven by

102 sin2(8)jo2(m)j1 2(m) fundamental; wj,wo2 (3.1)

1o2 cOS2(8)Jo2 (m)j 22 (m) 2nd harmonic; 2wl,2wo2 (3.2)

Io2cos2 (8)j64(m) 2nd-order intermodulation; w1* o (3.3)

o02sin 2(6)Jj1 2(M)j22(m) 3rd-order intermodulation; (3.4)2wj1-± o)2, 2wo2 ±- a),

where Io is the optical power incident on the detector, 8 is the modulator bias position, m is themodulation index, and Jn(m) is the Bessel function of the first kind of order n. The modulationindex is given by m = (7V2)(Vd/Vx), where Vd is the instantaneous drive voltage at wl and wo2,

32

and V., is the modulator half-wave voltage. For links that have bandwidths less than one octave,the SFDR is measured using the third-order intermodulation signals. The relative levels of thefundamental signal to the spur at the third-order intermodulation frequencies can be computedby taking the ratio of Eq. 3.4 to Eq. 3.1 given above. The ratio is constant for a specificmodulation index m and is independent of modulator bias 6. However, if the link is designed formultioctave bandwidth operation, the second-order intermodulaton signals must be inspected.The second-order intermodulation signals, ow ± 0)2, are larger than the second-order harmonics.The ratio of the fundamental to the second-order intermodulation signals is a function of themodulator bias point. Figure 3-5 shows the relative levels of Eq. 3.1 to 3.4 as a function ofphase bias error for a modulation index set to a value when the third-order products equal thenoise floor of the system shown in figure 3-2. The maximum allowable bias error is thencalculated based on the requirement that the SFDR not be degraded by the modulator biasing.

0

m -20 fl

cr -40LU f- +/- f2•0 . ---2 -- -- - -- - ---- 2.............-- ---

-120 2ff 2f-800 -100 2f..-.1-.-.

) -120 2f2 +-f 1"

cc -140

a.0 10 20 30 40 50 60 70 80 90

PHASE ERROR (deg)

Figure 3-5. Fundamental and spurious signal levels as a function of phase bias errorfor the minimum detectable modulation index.

The receiver bandwidth is an important factor in finding the sensitivity of the phase biaserror of the MZ modulators. For instance, if the receiver bandwidth is reduced, the system noisefloor is reduced and the spurious signals become more prevalent. In figure 3-5, the noise floorwould move lower and the second and third-order products would appear at a lower input drivelevel (smaller modulation index). Therefore, the output absolute noise floor level affects themeasurable spurious signal levels allowed, which determines the required modulator bias pointaccuracy.

33

The 30- to 500-MHz fiber optic link used for this measurement has an RF insertion loss of11 dB, an NF of 20 dB, and an SFDR of 109 dB/Hz23. The 2- to 18-GHz link has an RFinsertion loss of 27 dB, an NF of 35 dB, and a 108-dB/HzY SFDR for optimum modulatorbiasing. The effect of a 1-degree phase bias error on the SFDR for the 30- to 500-MHz link isshown in figure 3-6. The plot assumes a 30-kHz receiver bandwidth, which is appropriate forchannelized receiver or ultrawideband applications. It is already apparent that the SFDR of thislink is being severely limited by the harmonic levels. A similiar result is found for the 2- to18-GHz link. To emphasize this point further, a plot of the phase error tolerance (the second-order intermodulation signal level equals the third-order intermodulation level) versus receiverbandwidth for the lower frequency link is shown in figure 3-7. This curve suggests that formultioctave channelized receiver applications, a modulator phase bias error of 0.5 degree or lessis required. This level of modulator phase bias accuracy is extremely difficult to attain usingpassive modulator biasing techniques.

0 RF INSERTION LOSS = 11.6 dBE-20 V•t=4VE -20 -SFDR =80 dB1

-40 BANDWIDTH =30 kHzrr -40

' -60a -80 f2/o /-

W -100O -120 NOISE FLOOR Io -12,0 ,

"W /I-14 /U 140/ 2fj +/- f2 ,

-160 2f2 fl

-120 -100 -80 -60 -40 -20 0 20POWER APPLIED TO MODULATOR (dBm)

Figure 3-6. Effect of 1-degree modulator phase bias error on the 30-500-MHz fiber optic link.

A plot of the modulator phase bias drift versus temperature for the modulators used in the30- to 500-MI-Iz and 2- to 18-GHz links is shown in figure 3-8. Both the proton-exchanged andthe ion-diffused modulators show unacceptable bias point drifts over the 200 C to 1000 Ctemperature range. Even the most stable modulator designs will have difficulty maintaining thebias point to the degree required. Hence, active control of the modulator bias point is imperativefor high-performance multioctave link applications. Computer-controlled bias point stabilizationresults between 200 C and 100°C are shown in figure 3-9 for both links investigated. A secondharmonic suppression as high as 20 dB was obtained for both links, which translates intoincreased SFDR. Using the computer control, the bias points for both links were maintained to

34

MODULATION LEVEL (dBm)-35 -30 -25 -20 -15 -10 -5 0

2.5

2-J20

1.5 a12 R

cc

0 .5 Cl)

0~a10 Hz 10 kHz 10 MHz 10 GHz

RECEIVER BANDWIDTH (Hz)

Figure 3-7. Modulator phase bias tolerance as a function of receiverbandwidth for the 500-MHz fiber optic link.

60.0

40.0

"R 20.00 ION DIFFUSED,

0. 18 GHz BANDWIDTH

0 -20.0LU n PROTON/ -40.0 O< EXCHANGED,n -60.0 500 MHz BANDWIDTH

-80.0

-100.0 I :oi l !If) UO U) LO U) U*) U) LI)N C .J 113 n (D r t- co

TEMPERATURE (-C)

Figure 3-8. Modulator phase bias drift as a function of temperature forproton-exchanged and titanium-indiffused lithium niobate optical modulators.

35

-30.00

-35.00

9 -40.00

Z -4500OPEN LOOP0_ -45.00 /u)

o3-50.00

a.. - COMPUTERCONTROLLED

-60.00

-65.00 ,

-70.00

N I) U) 0 UO 0 tI 0 Ln 0 LnNY C Y V) IV "W n kn cc(0 NTEMPERATURE ('C)

(a) Ion-diffused modulator.

-30.00

-35.00

.-40.00Z0 - OPEN LOOPen -45.00Enw( -50.00D 0 COMPUTERU) .55.00 CONTROLLED

-60.00

-65.00 J ', 2 , ' , ' , : ' ' '" 'n 8 ) 0 L LO O o Lo Otno

TEMPERATURE ('C)

(b) Proton-exchanged modulator.

Figure 3-9. Computer-controlled modulator phase bias stabilization results as a functionof temperature for both low- and high-frequency fiber optic links.

36

within I degree, which was limited by the time consant of the feedback circuitry. Better biaspoint control should be possible using this technique along with optimized feedback algorithms.This should allow multioctave fiber optic links employing MZ optical modulators to approachthird-order intermodulation-limited SFDRs. This system improvement should enable EO EMEmonitoring systems to operate over the wide temperature ranges expected in the shipboardenvironment.

3.1.3 Remote Optical Polarization Control of EM Field Sensor

The performance of externally modulated remote fiber optic links similiar to that shown infigure 2-1 is subject to polarization fading if the proper polarization is not presented to themodulator. The use of PMF is a common method for overcoming this problem, and due to itsrelative maturity, it is the method to be used in the FY 94 shipboard demonstration for thisproject. However, PMF is difficult to use, is prone to damage, and can add significant cost to thesystem. Alternatively, two techniques have been investigated which eliminate the need for PMFin remote link applications. One technique uses a quasidepolarized optical source by combiningtwo orthogonally polarized light beams in a single-mode optical fiber.[22] In this case, themodulator ideally accepts the same amount of optical power irrespective of fiber polarizationeffects. With this technique, it is critical that the power of the two lasers is identical, that thepolarizations are truly orthogonal, and that the laser beat frequency is pushed beyond thefrequency range of interest. These are not trivial experimental tasks. Another technique relieson active polarization control with standard SMF to compensate for polarization drifts in thefiber between the source and the controller and in the fiber leading to the sensor. With thistechnique, much lower cost fiber can be used, which would permit simple field repairs and theuse of optic.i sources that are pigtailed with standard fiber rather than PMF. This polarization-control technique has been investigated for its application to the shipboard EME monitoringsystem.

Several polarization-controlling systems have been developed based on either lithiumniobate, rotating waveplates, fiber squeezers, or fiber cranks.[231 However, these systems areeither expensive, have high insertion loss, or mechanically fatigue the fiber, which can some-times result in breakage. Liquid crystals can be used for polarization control and do not sufferfrom these limitations.[24-] While response times of liquid crystals were once considered to betoo slow for some applications,[12 methods are being investigated to improve the response timeuntil it is practical for virtually all communication and sensor applications. This Sectiondescribes work demonstrating the use of a liquid-crystal polarization controller in a remote fiberoptic link configuration similiar to that used in the EO EME monitoring system.

The experimental setup is shown in figure 3-10. The source is a 10-mW Nd:YAG solid-statelaser emitting at 1.32 gm, followed by a quarter-wave plate and a half-wave plate. The wave-plates are mounted on stages which can be rotated to produce any desired polarization andeffectively simulate the effect of birefringence in standard SMF between the source and thepolarization controller. The liquid-crystal polarization controller consists